ИСТИНА |
Войти в систему Регистрация |
|
ИПМех РАН |
||
The criminal situation in the regions of the Russian Federation is the key for the need to study the parameters of a speech signal in relation to foreign language speech in order to conduct phonoscopic examinations in an ethnic language (Tajik, Gypsy, etc.). The need for an expert who is not a native speaker of the language under study to determine the key prosodic characteristics of foreign language speech, taking into account the perceptual-auditory and acoustic types of analysis of phonograms in order to identify a foreign-language speaker, poses a number of qualitatively new tasks for computer linguistics. Prosody plays a special role in personality identification as a fundamental paraverbal means of spoken speech communication. Despite the multifunctionality of prosody in spoken speech dialogic discourse, there are still no information sources that would give a structured idea of the stability and informativeness of the prosodic characteristics of discourse in a language that is unfamiliar to an expert. This paper describes a technique for identifying foreign-language speakers in Tajik and Gypsy languages by an expert who is a native speaker of the Russian language, by studying the prosodic organization of an utterance using perceptual-auditory and acoustic types of analysis based on the identification module as part of a specialized sound editor SIS II (developed by LLC Speech Technology Center, St. Petersburg), a universal acoustic-linguistic platform for the study of voice and speech, designed to automate the process of speaker identification.